On the Use of Two-Way Fixed Effects Models for Policy Evaluation During Pandemics

Fiche du document

Date

21 juin 2021

Type de document
Périmètre
Identifiant
Source

arXiv

Collection

arXiv

Organisation

Cornell University


Mots-clés

Economics - Econometrics

Sujets proches En

Pattern Model Assessment Use of

Citer ce document

Germain Gauthier, « On the Use of Two-Way Fixed Effects Models for Policy Evaluation During Pandemics », arXiv, ID : 10670/1.rmkquk


Métriques


Partage / Export

Résumé 0

In the context of the Covid-19 pandemic, multiple studies rely on two-way fixed effects (FE) models to assess the impact of mitigation policies on health outcomes. Building on the SIRD model of disease transmission, I show that FE models tend to be misspecified for three reasons. First, despite misleading common trends in the pre-treatment period, the parallel trends assumption generally does not hold. Second, heterogeneity in infection rates and infected populations across regions cannot be accounted for by region-specific fixed effects, nor by conditioning on observable time-varying confounders. Third, epidemiological theory predicts heterogeneous treatment effects across regions and over time. Via simulations, I find that the bias resulting from model misspecification can be substantial, in magnitude and sometimes in sign. Overall, my results caution against the use of FE models for mitigation policy evaluation.

document thumbnail

Par les mêmes auteurs

Sur les mêmes sujets

Sur les mêmes disciplines

Exporter en